Tao Xie

Meta’s New Breakthrough in Multilingual Speech AI! 🎙️ Meta AI has just released Omnilingual ASR the world’s first open-source speech recognition system supporting 1,600+ languages, including 500 that have never been covered by any ASR model before! 🔥 Built with a 7B-parameter encoder–decoder architecture and trained on 4.3M hours of multilingual audio, it brings speech technology closer to every community not just high-resource languages. 💡What makes it special: Recognizes 1,600+ languages Zero-shot support for unseen languages Open-source on GitHub check the live on Hugging Face https://lnkd.in/g8FZuBtU 🌏 A huge step toward inclusive AI and language preservation — making voice tech accessible to everyone, everywhere. #AI #SpeechRecognition #OmnilingualASR #MetaAI #MultilingualAI #OpenSource #ASR #LLM #Inclusion #HuggingFace

15

AI agents are hitting a new bottleneck: context memory. The newly renamed NVIDIA CMX context memory storage platform, based on the NVIDIA STX storage architecture announced at #NVIDIAGTC and powered by NVIDIA BlueField-4, creates a pod-level context tier built for KV cache with up to 5x higher TPS and 5x better power efficiency. Explore how CMX unlocks scalable, AI-native context memory for agentic inference. https://bit.ly/4rKs76h

3

Deploy Kimi-K2-Instruct, a frontier mixture-of-experts model, on Lambda using vLLM in four steps: 1️⃣ Spin up an NVIDIA Blackwell instance on Lambda. 2️⃣ Start a vLLM deployment server with a single setup block. 3️⃣ Run a reproducible benchmark against that server. 4️⃣ Share the exact model and image references to reproduce. Read the technical blog to get started 🔗

1

Just watched the Cannes Lions interview with Mustafa Suleyman and Chloe Shih—and it’s packed with insights for anyone navigating the evolving AI landscape. As someone who works closely with customers, I’m always thinking about how we build AI that’s not just powerful, but practical, human-centered, and trustworthy. This conversation hit all the right notes.  Key takeaway? The 4Qs of AI—a framework I’ll be referencing often: ➡️ IQ – Intelligence is the foundation ➡️ EQ – Emotional awareness builds trust ➡️ AQ – Adaptability keeps us relevant ➡️ SQ – Social intelligence fosters collaboration This isn’t just a framework for AI—it’s a leadership mindset. Mustafa also shared how Copilot is evolving to amplify human creativity, not replace it. That’s a vision I can get behind—and one I see resonating with customers every day. From Beatles references to a surprisingly relatable Copilot moment, there’s great advice here for the next generation of builders! Let’s keep building AI that’s smart and socially aware. #AI #Leadership #MicrosoftAI #Copilot #CustomerInnovation #4Qs #CannesLions2025 #TechForGood

22

Over the past few months I've been trying to get deeper into RTL and understand how real hardware blocks inside processors are built. One thing that caught my attention while reading about modern CPUs and GPUs from companies like AMD and Nvidia is how much they rely on hardware performance monitoring. These chips have internal counters that track different micro-architectural events so engineers can analyze how the processor behaves under real workloads. That idea was really interesting to me, so I decided to try building a simplified version of a Performance Monitoring Unit myself. I implemented a parameterized PMU in SystemVerilog that monitors multiple event signals and tracks them using independent saturating counters. The design is modular, scalable, and verified through simulation with assertions and a testbench. This project was a great learning experience for me because it pushed me to think more about how hardware blocks are structured, not just how to write Verilog that compiles. Still a lot more to learn in RTL and silicon design, but building things like this is honestly the most fun part of studying computer engineering. Repo if anyone wants to check it out: https://lnkd.in/ejRKvtcm

58

3 Comments

AI agents are hitting a new bottleneck: context memory. The newly renamed NVIDIA CMX context memory storage platform, based on the NVIDIA STX storage architecture announced at #NVIDIAGTC and powered by NVIDIA BlueField-4, creates a pod-level context tier built for KV cache with up to 5x higher TPS and 5x better power efficiency. Explore how CMX unlocks scalable, AI-native context memory for agentic inference. https://bit.ly/4brZynR

4

Introducing the M5 The M5 chip brings significant advancements in performance and efficiency across various applications. 1. Massive GPU / AI performance boost The M5 features a next-generation 10-core GPU architecture, with each core equipped with an integrated Neural Accelerator. This results in over 4× the peak GPU compute performance for AI tasks compared to the M4. 2. Stronger CPU and enhanced Neural Engine With up to a 10-core CPU (6 efficiency cores + 4 performance cores), the M5 offers approximately 15% faster multithreaded performance over the M4. Additionally, the Neural Engine has been upgraded to 16 cores, enhancing performance for on-device AI and intelligent features. 3. Higher memory bandwidth / unified memory improvements The M5 provides around 153 GB/s of unified memory bandwidth, which is approximately 30% more than the M4. This unified memory architecture allows all components (CPU, GPU, Neural Engine) to access a larger memory pool, enabling larger AI models to run entirely on-device. 4. Graphics & rendering enhancements The improvements in the GPU also boost graphics performance, including ray tracing capabilities. Apple claims up to a ~45% uplift compared to the M4 in ray-traced workloads, enhancing visual fidelity, rendering speed, and smoothness in demanding 3D and graphics applications. 5. Deployment in new hardware + energy efficiency / environmental goals The M5 will power new devices, including the 14-inch MacBook Pro, iPad Pro, and upgraded Vision Pro, all of which are available for pre-order.

7